Fragility assessment of tunnels in soft soils using artificial neural networks

Recent earthquakes have shown that tunnels are prone to damage, posing a major threat safety and having cascading socioeconomic impacts. Therefore, reliable models needed for the seismic fragility assessment of underground structures quantitative evaluation expected losses. Based on previous researches, this paper presented probabilistic framework based an artificial neural network (ANN), aiming at development curves circular in soft soils. Initially, two-dimensional incremental dynamic analysis nonlinear soil-tunnel system was performed estimate response tunnel under ground shaking. The effects soil-structure-interaction motion characteristics were adequately considered within proposed framework. An ANN employed develop demand model, its results compared with traditional linear regression models. Fragility generated various damage states, accounting associated uncertainties. indicate ANN-based can models, similar capabilities as approaches, lower computational cost is required. be adopted risk typical soils subjected loading, they facilitate decision-making management toward more resilient transport infrastructure.